BEGIN:VCALENDAR VERSION:2.0 PRODID:-//CERN//INDICO//EN BEGIN:VEVENT SUMMARY:Towards mass composition study with KASCADE using deep learning DTSTART:20220203T143500Z DTEND:20220203T145500Z DTSTAMP:20260722T082500Z UID:indico-contribution-7986@events.icecube.wisc.edu DESCRIPTION:Speakers: N. Petrov (Novosibirsk State University / Budker Ins titute of Nuclear Physics)\, V. Lenok (Karlsruhe Institute of Technology\, Institute for Astroparticle Physics)\, O. Shchegolev (Moscow Institute of Physics and Technology / Institute for Nuclear Research of the Russian Ac ademy of Sciences)\, V. Tokareva (Karlsruhe Institute of Technology\, Inst itute for Astroparticle Physics)\, Dmitriy Kostunin (DESY)\, P. Bezyazeeko v (Applied Physics Institute\, Irkutsk State University)\, S. Golovachev ( JetBrains Research)\, Margarita Tsobenko (Higher School of Economics Unive rsity - St. Petersburg)\, Vladimir Sotnikov (Jetbrains)\, I. Plokhikh (Nov osibirsk State University / Institute of Thermophysics SB RAS)\, Daniil R eutsky (Moscow Institute of Physics and Technology)\n\nKASCADE was an ai r-shower detector located in Karlsruhe Institute of Technology. It consist ed of scintillating detectors which were arranged in a 16×16 grid and rec orded signals from secondary particles of air-showers. Data has been acqui red from 1996 till 2013 and then has been made available online. Our goal is to find out\, whether we can accurately reconstruct the initial particl e by that data from the ground level. At the current stage of our work we use CORSIKA simulations of this experiment\, getting data for 5 mass group s of particles and training our classifiers on them. We have tested two mo dels: decision trees and convolutional neural network. After the training step we apply our models to the data from the real KASCADE experiment and check the credibility of the predicted particles distribution. Contrary to decision trees\, the CNN are more sensitive to irregularities in the raw data and thus the data have to be preprocessed in term of application of a dditional quality cuts. In this talk we present the performance of develop ed classifiers and show our progress in preparation of the raw data for CN N.\n\nhttps://events.icecube.wisc.edu/event/141/contributions/7986/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7986/ END:VEVENT BEGIN:VEVENT SUMMARY:Machine learning based event reconstruction in Telescope Array sur face detector DTSTART:20220202T140000Z DTEND:20220202T144500Z DTSTAMP:20260722T082500Z UID:indico-contribution-7987@events.icecube.wisc.edu DESCRIPTION:Speakers: Oleg Kalashev (INR RAS Moscow)\n\nThe surface detect or of the Telescope Array (TA) experiment is the largest one in the northe rn hemisphere. We overview the machine learning based event reconstruction methods being developed by the TA collaboration. The key idea is to use f ull detector Monte Carlo simulation to obtain the raw detector signal as a function of the primary particle properties and to train deep convolution al neural network to model the inverse function. The above technique can b e used to enhance the energy and arrival direction reconstruction for the individual events and to estimate the mass composition for an ensemble of events.\n\nhttps://events.icecube.wisc.edu/event/141/contributions/7987/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7987/ END:VEVENT BEGIN:VEVENT SUMMARY:Search for optimal deep neural network architecture for gamma dete ction at KASCADE DTSTART:20220203T160000Z DTEND:20220203T162000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7993@events.icecube.wisc.edu DESCRIPTION:Speakers: S. Golovachev (JetBrains Research)\, O. Shchegolev ( Moscow Institute of Physics and Technology / Institute for Nuclear Researc h of the Russian Academy of Sciences)\, Daniil Reutsky (Moscow Institute of Physics and Technology)\, Vladimir Sotnikov (Jetbrains)\, V. Tokareva (Karlsruhe Institute of Technology\, Institute for Astroparticle Physics)\ , V. Lenok (Karlsruhe Institute of Technology\, Institute for Astroparticl e Physics)\, P. Bezyazeekov (Applied Physics Institute\, Irkutsk State Uni versity)\, Dmitriy Kostunin (DESY)\, I. Plokhikh (Novosibirsk State Univer sity / Institute of Thermophysics SB RAS)\, N. Petrov (Novosibirsk State U niversity / Budker Institute of Nuclear Physics)\, Margarita Tsobenko (Hig her School of Economics University - St. Petersburg)\n\nWe focus on the no vel data analysis from KASCADE\, one of the most successful cosmic ray det ectors in the >PeV range. The detector operated for about 15 years\, its d ata are publicly accessible. The data archive includes about half a billio n recorded air showers. Extensive air showers generated by ultrahigh-energ y gamma-rays (not detected at the moment) are of particular research inter est\, since information about particles of this type allows us to learn ab out the properties of cosmic ray sources\, as well as to study the nature of diffuse photons. The main problem is that this type of particle is diff icult to distinguish against the background of cosmic protons\, since the signatures left by protons and photons have similar characteristics. To so lve this problem\, we present a primary particle type classifier (gamma or proton) trained on the basis of the simulation data of the KASCADE detect or. For classification\, various approaches are applied using deep learnin g methods.\n\nhttps://events.icecube.wisc.edu/event/141/contributions/7993 / LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7993/ END:VEVENT BEGIN:VEVENT SUMMARY:Open questions in deep learning techniques for the radio detection DTSTART:20220203T190000Z DTEND:20220203T194500Z DTSTAMP:20260722T082500Z UID:indico-contribution-7982@events.icecube.wisc.edu DESCRIPTION:Speakers: Dmitriy Kostunin (DESY)\n\nNowadays the deep learnin g techniques are broadly applied for the processing of radio signals gener ated in air-showers. The majority of the implementations are based on the convolutional neural networks (CNN) running of 1D arrays containing finite waveforms with radio impulses. This approach has shown its feasibility an d is able to be implemented for the both trigger- and high- levels of data collection and analysis. However there is a room for the improvement and some open questions. During my talk we reviewed the current progress in th e field\, pointed the important issues and their possible solutions\, and shared and discussed ideas of the optimal application of this technique.\n \nhttps://events.icecube.wisc.edu/event/141/contributions/7982/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7982/ END:VEVENT BEGIN:VEVENT SUMMARY:Exploitation of Symmetries and Domain Knowledge in Deep Learning A rchitectures DTSTART:20220201T160000Z DTEND:20220201T164500Z DTSTAMP:20260722T082500Z UID:indico-contribution-7984@events.icecube.wisc.edu DESCRIPTION:Speakers: Mirco Huennefeld (Universität Dortmund)\, IceCube C ollaboration\n\nThe field of deep learning has become increasingly importa nt for particle physics experiments\, yielding a multitude of advances\, p redominantly in event classification and reconstruction tasks. Many of the se applications have been adopted from other domains. However\, data in th e field of physics are unique in the context of machine learning\, insofar as their generation process and the laws and symmetries they abide by are usually well understood. Most commonly used deep learning architectures f ail at utilizing this available domain knowledge. \nIn this contribution\, the importance of utilizing domain knowledge is highlighted and a hybrid reconstruction method is introduced that combines the benefits of maximum- likelihood estimation with those of deep learning. Domain knowledge\, such as invariances and detector characteristics\, can easily be incorporated in this approach. Although applicable to any simulation based experiment\, the hybrid method is illustrated by the example of event reconstruction i n IceCube.\n\nhttps://events.icecube.wisc.edu/event/141/contributions/7984 / LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7984/ END:VEVENT BEGIN:VEVENT SUMMARY:Deep Learning for Classification and Denoising of Cosmic-Ray Radio Signals DTSTART:20220203T194500Z DTEND:20220203T201500Z DTSTAMP:20260722T082500Z UID:indico-contribution-7988@events.icecube.wisc.edu DESCRIPTION:Speakers: Frank Schroeder (University of Delaware / Karlsruhe Institute of Technology)\, Abdul Rehman (University of Delaware)\, Alan Co leman (University of Delaware)\n\nRadio emission\, produced mainly as a re sult of the geomagnetic deflection of oppositely charged particles within the cosmic-ray air showers\, is contaminated by backgrounds such as the co ntinuous Galactic background and thermal noise. This irreducible backgroun d poses a significant challenge for radio detection of air showers. To mit igate this effect of background we employ machine learning (ML) techniques . These techniques such as convolutional neural networks (CNNs) have been widely used to analyze visual imagery. It is only recently that these tech niques have been adopted in many fields of science for the purpose of reco gnizing different patterns in the data. In this work\, we use CNNs with th e following two goals: to classify waveforms with signals against those th at include only noise and to extract the underlying radio signals from the contaminated traces. To produce the required dataset for training the mod els\, we use CoREAS simulations which calculate the radio signals from air showers. For background we considered Cane Model for average Galactic noi se\, with an additional thermal component. Both signal and background trac es are filtered in the 50 - 350 MHz frequency band before training. With t hese ML models\, we aim to improve the detection threshold and also the re construction efficiency of the radio technique for cosmic-ray air showers. \n\nhttps://events.icecube.wisc.edu/event/141/contributions/7988/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7988/ END:VEVENT BEGIN:VEVENT SUMMARY:Composition Analysis of cosmic-rays at IceCube Observatory\, using Graph Neural Networks DTSTART:20220201T164500Z DTEND:20220201T173000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7989@events.icecube.wisc.edu DESCRIPTION:Speakers: IceCube Collaboration\, Paras Koundal (Karlsruhe Ins titute of Technology)\n\nThe IceCube Neutrino Observatory\, located at the South Pole\, is a multi-component detector that detects high-energy parti cles from astrophysical sources. Cosmic Rays (CRs) are charged particles f rom these astrophysical accelerators. CRs and CR-induced air-showers furni sh us with the possibility to discern the fundamental properties and behav ior of such sources. When coupled to the IceTop surface array\, IceCube a ffords unique three-dimensional detection and cosmic-ray analysis in the t ransition region from galactic to extragalactic sources. This work tries t o improve the estimation of CR primary mass on a per-event basis in the me ntioned energy range. The work benefits from using the full in-ice shower footprint and additional composition-sensitive air-shower parameters\, in addition to global shower-footprint parameters already used in an earlier work. A Graph Neural Network (GNN) based implementation uses the full in-i ce shower footprint. Described using nodes and edges\, graphs allow us to efficiently represent relational data and learn hidden representations of input data to obtain better model accuracy. Mapping in-ice IceCube detecto rs\, DOMs(Digital Optical Module)\, as a graph emerges as a natural soluti on. Using GNNs for cosmic-ray analysis at IceCube also has the added benef it of allowing an easier re-implementation to the planned next-generation upgraded instrument\, called IceCube-Gen2.\n\nhttps://events.icecube.wisc. edu/event/141/contributions/7989/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7989/ END:VEVENT BEGIN:VEVENT SUMMARY:Photon flux calculation using Deep Learning DTSTART:20220203T162000Z DTEND:20220203T165500Z DTSTAMP:20260722T082500Z UID:indico-contribution-7991@events.icecube.wisc.edu DESCRIPTION:Speakers: Jigar Bhanderi\, Dmitry Malyshev\n\nOptical interfe rometry provides a sub-milliarcsecond resolution of astronomical objects. Intensity interferometry is a part of optical interferometry\, which deals with correlation of intensities rather than amplitude of waves. For succe ssful measurements\, one needs large collecting area\, such as an array of several telescopes separated by hundreds of meters with good time resolut ion of photon flux\, e.g\, imaging atmospheric Cherenkov telescopes such a s H.E.S.S and CTA. The measurements have high photon rates\, so that the p ulses in PMTs from individual photons overlap. As a result\, the rate dete rmination by counting is unfeasible. We use several neural networks (such as CNNs\, LSTMs\, GRUs) in order to determine the rate of photons detected by the PMTs.\n\nhttps://events.icecube.wisc.edu/event/141/contributions/7 991/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7991/ END:VEVENT BEGIN:VEVENT SUMMARY:Air shower reconstruction using a Graph Neural Network for the Ice Act telescopes DTSTART:20220201T193000Z DTEND:20220201T200000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7992@events.icecube.wisc.edu DESCRIPTION:Speakers: Larissa Paul (Marquette University)\, The IceCube Co llaboration\, Thomas Bretz (RWTH Aachen University)\, John Hewitt (Univers ity of North Florida)\, Adrian Zink (Friedrich-Alexander-Universität Erla ngen)\n\nThe IceAct telescopes are prototype Imaging Air Cherenkov telesco pes (IACTs) situated at the IceCube Neutrino Observatory at the geographi c South Pole. The telescopes camera consist of 61 silicon photomultipliers  (SiPMs) with a hexagonal light guide glued to each SiPM. The IceAct tel escopes measure the electromagnetic air shower component of cosmic rays in the atmosphere\, which is complementary to the muonic component measure d by the IceCube in-ice detector and the particle footprint measured  at the surface by IceTop.  The shape of the events and the number of S iPMs hit per event within the IceAct telescopes\, and the possibility of c ombining information from different detector components\, makes the IceAct data a perfect candidate for a reconstruction of particle type and energy  using a graph neural network (gnn). In contrast to other neural network s\, gnns do not need a fixed structure between the nodes\, the number node s can differ between events and the connection between the nodes can be de fined individually for each pair of nodes. A Monte Carlo study for a first gnn reconstruction of air shower events with the IceAct telescopes will b e presented.\n\nhttps://events.icecube.wisc.edu/event/141/contributions/79 92/ LOCATION:Embassy Suites by Hilton Newark Wilmington South URL:https://events.icecube.wisc.edu/event/141/contributions/7992/ END:VEVENT BEGIN:VEVENT SUMMARY:Energy Reconstruction with Convolutional Neural Networks in IceTop DTSTART:20220201T220000Z DTEND:20220201T223000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7997@events.icecube.wisc.edu DESCRIPTION:Speakers: Frank McNally (Mercer University)\, The IceCube Coll aboration\n\nIceTop\, the surface component of the IceCube Neutrino Observ atory\, consists of 81 stations that detect air showers produced by cosmic ray interactions with the atmosphere. An accurate energy estimator for Ic eTop is essential for studying the nature of the cosmic ray spectrum aroun d the knee (300 TeV - 1 EeV). Using over 400\,000 simulated events\, we tr ained an array of convolutional deep neural networks (CNNs) to reconstruct the energy of a cosmic ray primary based on the charges detected at the s urface. Preliminary results show that charge-only CNN models can deliver a n energy resolution better than 10%\, with significant improvements when i ncluding reconstructed zenith. This result is consistent with independent energy reconstructions used by IceCube\, and indicates the promise of a de ep-learning approach.\n\nhttps://events.icecube.wisc.edu/event/141/contrib utions/7997/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7997/ END:VEVENT BEGIN:VEVENT SUMMARY:What slow down cosmic ray analysis and what can we do about them? DTSTART:20220203T213000Z DTEND:20220203T215500Z DTSTAMP:20260722T082500Z UID:indico-contribution-7994@events.icecube.wisc.edu DESCRIPTION:Speakers: Xinhua Bai (South Dakota School of Mines and Technol ogy)\n\nCosmic ray analysis relies on multiple steps including calibration \, simulation\, event reconstruction and interpretation\, etc. Because of their broad energy coverage and sophisticated analysis and simulation tech niques\, large cosmic ray projects often suffer from their science analysi s falling behind data collection. This challenge may be more severe in nex t generation multi-messenger astroparticle physics projects in which more hybrid detection techniques will be used. This presentation will list a fe w key "nodes" that often slows down the analysis and excite a roundtable d iscussion to see how we can mitigate the challenge by harnessing Big Data revolution.\n\nhttps://events.icecube.wisc.edu/event/141/contributions/799 4/ LOCATION:Embassy Suites by Hilton Newark Wilmington South URL:https://events.icecube.wisc.edu/event/141/contributions/7994/ END:VEVENT BEGIN:VEVENT SUMMARY:Crowdsourcing your training labels with Zooniverse DTSTART:20220203T210000Z DTEND:20220203T213000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7995@events.icecube.wisc.edu DESCRIPTION:Speakers: Lucy Fortson (University of Minnesota)\n\nIn this pr esentation\, I will describe the Zooniverse.org citizen science platform a s a tool to gather labels from over 2.5 million dedicated volunteers world wide who are motivated to participate in scientific research. Hundreds of research teams now turn to Zooniverse for crowdsourcing tasks such as imag e classification and annotation which provide the large labeled data sets needed for optimal training of machine algorithms. I will provide examples from across several relevant domains including particle physics\, multi-m essenger astrophysics and IACT event categorization\, with a focus on the Muon Hunter project used to gather millions of labels to train a CNN for a n IACT calibration pipeline. I will demonstrate the ease with which a proj ect can be developed with the Zooniverse Project Builder tools and describ e the infrastructure available for integrating machine learning with Zooni verse including sophisticated active learning techniques.\n\nhttps://event s.icecube.wisc.edu/event/141/contributions/7995/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7995/ END:VEVENT BEGIN:VEVENT SUMMARY:Improving the gamma-hadron separation for air showers at the IceCu be Neutrino Observatory DTSTART:20220203T165500Z DTEND:20220203T173000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7976@events.icecube.wisc.edu DESCRIPTION:Speakers: Federico Bontempo (Karlsruhe Institute of Technology )\, IceCube collaboration\n\nThe IceCube Neutrino Observatory is a unique experiment located at the geographic South Pole. It is composed of two det ectors: an optical array deep in the ice and an array of ice-Cerenkov tank s at the surface called IceTop. The combination of the two detectors can b e exploited for the study of cosmic rays and the search for PeV photons. I n particular\, the in-ice detector measures the high-energy muonic compone nt of air showers\, and the surface detector all shower component and can be used for the general shower reconstruction. The aim of this work in pro gress is to discriminate between photon initiated and cosmic ray initiated air showers. This discrimination is performed using a machine learning te chnique named Random Forest. This is a supervised machine learning techniq ue that predicts unknown data after studying labeled data. The physics qua ntities used for this study are the charges measured by the in-ice detecto r\, the zenith angle\, a parameter that describes the in ice containment o f the shower\, the reconstructed energy and a likelihood estimator that ca ptures both the presence of individual muons and charge fluctuations in th e surface array. \nFurthermore\, the planned enhancement of IceTop\, compr ised of surface radio antennas and scintillator panels\, will contribute t o the improvement of the gamma-hadron separation.\n\nhttps://events.icecub e.wisc.edu/event/141/contributions/7976/ LOCATION:Embassy Suites by Hilton Newark Wilmington South URL:https://events.icecube.wisc.edu/event/141/contributions/7976/ END:VEVENT BEGIN:VEVENT SUMMARY:Cosmic rays primary energy estimation using Machine Learning and c ombined reconstruction DTSTART:20220201T213000Z DTEND:20220201T220000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7985@events.icecube.wisc.edu DESCRIPTION:Speakers: Diana Leon Silverio (South Dakota School of Mines an d Technology)\, Matthias Plum (Marquette University)\, Xinhua Bai (South D akota School of Mines and Technology)\, IceCube Collaboration\n\nThe IceCu be Neutrino Observatory at the South Pole is capable of measuring two comp onents of the cosmic rays air shower. The electromagnetic component using a km2 surface array IceTop\, and the high-energy muonic component using km 3 in-ice array IceCube between 1.5 and 2.5 km below the surface. The combi nation of both arrays in conjunction with a new flexible curvature and new timing fluctuation function provides an opportunity for possible improvem ents of cosmic rays reconstruction. This work presents a preliminary inves tigation of possible improvements of cosmic rays primary energy estimation (proton\, iron\, helium\, and oxygen) by using Machine Learning technique s and combined reconstruction.\n\nhttps://events.icecube.wisc.edu/event/14 1/contributions/7985/ LOCATION:Embassy Suites by Hilton Newark Wilmington South URL:https://events.icecube.wisc.edu/event/141/contributions/7985/ END:VEVENT BEGIN:VEVENT SUMMARY:Cosmic ray mass composition study using a Random Forest applied to data from the IceAct telescopes DTSTART:20220201T200000Z DTEND:20220201T201500Z DTSTAMP:20260722T082500Z UID:indico-contribution-7998@events.icecube.wisc.edu DESCRIPTION:Speakers: Adrian Zink (Friedrich-Alexander-Universität Erlang en)\, John Hewitt (University of North Florida)\, Thomas Bretz (RWTH Aache n University)\, The IceCube Collaboration\, Larissa Paul (Marquette Univer sity)\, Matthias Plum (Marquette University)\, Karen Andeen (Marquette Uni versity)\n\nThe IceAct telescopes are prototype Imaging Air Cherenkov tele scopes (IACTs) situated at the IceCube Neutrino Observatory at the geogra phic South Pole. The IceAct telescopes measure the electromagnetic air sho wer component of cosmic rays in the atmosphere\, which is complementary t o the muonic component measured by the IceCube in-ice detector and the particle footprint measured at the surface by IceTop. For this Monte Carlo study a random forest is used to analyze the mass composition of the cosm ic rays spectrum using the three independent measurements of the cosmic ra y air showers provided by the different detector components.\n\nhttps://ev ents.icecube.wisc.edu/event/141/contributions/7998/ LOCATION:Embassy Suites by Hilton Newark Wilmington South URL:https://events.icecube.wisc.edu/event/141/contributions/7998/ END:VEVENT BEGIN:VEVENT SUMMARY:Machine Learning for High-Energy Physics Reconstruction and Analys is DTSTART:20220201T150000Z DTEND:20220201T153000Z DTSTAMP:20260722T082500Z UID:indico-contribution-8000@events.icecube.wisc.edu DESCRIPTION:Speakers: Sergei Gleyzer\n\nThe Large Hadron Collider (LHC) is delivering the highest energy proton-proton collisions ever recorded in t he laboratory\, permitting a detailed exploration of elementary particle p hysics at the highest energy frontier. In this talk\, I will discuss the a pplication of machine learning to problems in high-energy physics\, with a focus on the challenges associated with large\, complex datasets from the Large Hadron Collider\, and those expected from the High-Luminosity Large Hadron Collider. I will discuss the application of state-of-the-art machi ne learning methods to new physics searches at the LHC\, detector reconstr uction\, event simulation and real-time event filtering at the LHC.\n\nhtt ps://events.icecube.wisc.edu/event/141/contributions/8000/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/8000/ END:VEVENT BEGIN:VEVENT SUMMARY:Deep Learning for Air Shower Reconstruction at the Pierre Auger Ob servatory DTSTART:20220202T160000Z DTEND:20220202T163000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7977@events.icecube.wisc.edu DESCRIPTION:Speakers: for the Pierre Auger Collaboration\, Jonas Glombitza (RWTH AACHEN UNIVERSITY)\n\nThe measurement of the mass composition of ul tra-high energy cosmic rays constitutes one of the biggest challenges in a stroparticle physics. Detailed information on the composition can be obtai ned from measurements of the depth of maximum of air showers\, Xmax\, with the use of fluorescence telescopes\, which can be operated only during cl ear and moonless nights.\n\nUsing deep neural networks\, it is now possibl e for the first time to perform an event-by-event reconstruction of Xmax w ith the Surface Detector (SD) of the Pierre Auger Observatory. Therefore\, previously recorded data can be analyzed for information on Xmax\, and th us the cosmic-ray composition. Since the SD operates with a duty cycle of almost 100% and its event selection is less strict than for the Fluorescen ce Detector (FD)\, the gain in statistics with respect to the FD is almost a factor of 15 for energies above $10^{19.5}$ eV.\n\nIn this contribution \, we introduce the neural network particularly designed for the SD of the Pierre Auger Observatory. We evaluate its performance using three differe nt hadronic interaction models and verify its functionality using Auger hy brid measurements. \nFinally\, we quantify the expected systematic uncerta inties and show that the method permits to determine the first two moments of the Xmax distributions up to the highest energies.\n\nhttps://events.i cecube.wisc.edu/event/141/contributions/7977/ LOCATION:Embassy Suites by Hilton Newark Wilmington South URL:https://events.icecube.wisc.edu/event/141/contributions/7977/ END:VEVENT BEGIN:VEVENT SUMMARY:IACT event reconstruction with deep learning: some progress\, less ons learned\, and outlook from CTLearn DTSTART:20220203T145500Z DTEND:20220203T153000Z DTSTAMP:20260722T082500Z UID:indico-contribution-8002@events.icecube.wisc.edu DESCRIPTION:Speakers: Daniel Nieto (Instituto de Física de Partículas y del Cosmos and Departamento de EMFTEL\, Universidad Complutense de Madrid) \, Tjark Miener (Instituto de Física de Partículas y del Cosmos and Depa rtamento de EMFTEL\, Universidad Complutense de Madrid)\n\nCTLearn is a pr oject that aims at IACT event reconstruction through the usage deep-learni ng models. The associated software packages include modules for loading an d manipulating IACT data\, and handling the training and test of deep-lear ning architectures with TensorFlow\, using pixel-wise camera data as input . In this contribution we will comment on the challeges we faced so far\, the lessons learned\, our latest results\, and our plans for the future.\n \nhttps://events.icecube.wisc.edu/event/141/contributions/8002/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/8002/ END:VEVENT BEGIN:VEVENT SUMMARY:CORSIKA 8: A modern framework for high-energy cascade simulations DTSTART:20220202T194500Z DTEND:20220202T203000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7980@events.icecube.wisc.edu DESCRIPTION:Speakers: CORSIKA 8 Collaboration\, Remy Prechelt (University of Hawai'i)\n\nThe proliferation of innovative next-generation cosmic ray and neutrino observatories\, with unique geometries (Earth-skimming\, orbi tal\, in-ice\, etc.)\, and detection techniques (Cherenkov\, radio\, radar \, etc.)\, requires the simulation of ultrahigh energy particle cascades w hich are *challenging*\, if not *impossible*\, to perform with current sim ulation tools like CORSIKA 7 and AIRES. These existing codes\, which have been developed in FORTRAN for more than three decades\, can be challenging to extend or extensively modify due to their fundamental software archite cture\, as well as due to their rigid assumptions about event geometries\, the cascade environment\, and the underlying physics models.\n\nCORSIKA 8 is a *completely new simulation framework*\, developed in modern C++ from the ground up\, and is designed to perform high- and ultrahigh-energy par ticle cascades in matter. CORSIKA 8 has been designed to be extremely flex ible\, extensible\, and easy-to-use while also being extremely performant via the use of compile-time optimization and HPC techniques like SIMD\, pa rallelization\, and GPU acceleration. In particular\, CORSIKA 8 provides s tandard *"pluggable"* components for creating cascade simulations with uni que geometries and physics\, not only in air\, but also in any other media including water\, ice\, and the lunar regolith.\n\nWe present the current status of the CORSIKA 8 project including the currently supported hadroni c and electromagnetic physics models\, the included radio & Cherenkov emis sion modelling\, and give an introduction to new simulations that will be or are already enabled by the CORSIKA 8 project.\n\nhttps://events.icecube .wisc.edu/event/141/contributions/7980/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7980/ END:VEVENT BEGIN:VEVENT SUMMARY:CORSIKA and CONEX for air shower simulations DTSTART:20220202T190000Z DTEND:20220202T194500Z DTSTAMP:20260722T082500Z UID:indico-contribution-8001@events.icecube.wisc.edu DESCRIPTION:Speakers: Tanguy Pierog (Karlsruhe Institute of Technology (KI T)\, IAP)\n\nIn order to properly train neural networks to analyze air sho wer data\, it is necessary to have accurate simulations providing the nece ssary level of details required to extract the required information. The m ost popular tool is certainly the current version of CORSIKA and its fast option for 1D simulation CONEX. We will present the basic principles of th ese tools and how to use them properly. The limitations\, mostly coming fr om the hadronic interaction models\, will be addressed to avoid any over i nterpretation of what the simulations can really do.\n\nhttps://events.ice cube.wisc.edu/event/141/contributions/8001/ LOCATION:Embassy Suites by Hilton Newark Wilmington South URL:https://events.icecube.wisc.edu/event/141/contributions/8001/ END:VEVENT BEGIN:VEVENT SUMMARY:Good Bye DTSTART:20220203T222000Z DTEND:20220203T223000Z DTSTAMP:20260722T082500Z UID:indico-contribution-8044@events.icecube.wisc.edu DESCRIPTION:Speakers: Frank Schroeder (University of Delaware / Karlsruhe Institute of Technology)\n\nhttps://events.icecube.wisc.edu/event/141/cont ributions/8044/ LOCATION:Embassy Suites by Hilton Newark Wilmington South URL:https://events.icecube.wisc.edu/event/141/contributions/8044/ END:VEVENT BEGIN:VEVENT SUMMARY:Pattern Recognition for Multiple Interactions in a Neutron Monitor DTSTART:20220201T201500Z DTEND:20220201T203000Z DTSTAMP:20260722T082500Z UID:indico-contribution-8055@events.icecube.wisc.edu DESCRIPTION:Speakers: Alejandro Sáiz (Mahidol University)\, Achara Seripi enlert (National Astronomical Research Institute of Thailand (NARIT)\, Chi ang Mai 50180\, Thailand)\, Waraporn Nuntiyakul (Chiang Mai University)\, P.-S. Mangeard (University of Delaware)\, Suruj Seunarine (University of W isconsin–River Falls)\, David Ruffolo (Mahidol University)\, J. Clem (Un iversity of Delaware)\, Paul Evenson (University of Delaware)\n\nThe flux of Galactic cosmic rays at Earth is modulated by the long term magnetic va riations of the Sun (11-year sunspot cycle and 22-year magnetic solar cycl e). This process known as Solar modulation is most pronounced at 1 GeV and below. However\, it also operates at much higher energy\, still exhibitin g solar magnetic polarity dependence. For the last decades\, ground-based neutron monitors provided valuable observations of the solar modulation up to a rigidity cutoff of about 17 GV. To extend the energy range of the ne utron monitor observations\, we recently upgraded the electronics of the P rincess Sirindhorn Neutron Monitor in Thailand (PSNM\, the operating neutr on monitor at the highest geomagnetic rigidity cutoff) to record complex c ombinations of hits in multiple proportional counters. The variety of even t topology recorded at the PSNM indicates multiple sources: energetic atmo spheric nucleons (GeV-range)\, coincidence of secondary particles\, and po ssibly small air-shower core passing through the detector. We discuss thes e observations with a preliminary analysis of a detailed Monte-Carlo simul ation of energetic neutrons interacting in the detector.\n\nhttps://events .icecube.wisc.edu/event/141/contributions/8055/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/8055/ END:VEVENT BEGIN:VEVENT SUMMARY:State-of-art deep learning technologies and their application to a ir-shower reconstruction DTSTART:20220202T144500Z DTEND:20220202T153000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7975@events.icecube.wisc.edu DESCRIPTION:Speakers: Vladimir Sotnikov (JetBrains Research)\n\nOnce again \, the last several years reshaped the state-of-the-art in Computer Vision (CV). Non-convolutional approaches\, such as Vision Transformers (ViT) an d self-attention multi-layer perceptrons (SA-MLP)\, are quickly emerging\, combined with novel optimization techniques and pre-training methods. Not e that ViTs and SA-MLPs are evidently better at incorporating global infor mation about the input data\, they're also not spatially invariant\, which is more appropriate for the cosmic-ray air-showers detectors. This contri bution covers multiple approaches for the unsupervised pre-training - a te chnique that allows making model learn on the unlabeled (i.e.\, experiment al) data and thus increases the model performance. However\, each of the e xamined approaches is nontrivial to apply to air-showers\, which poses a c hallenge yet to be solved.\n\nhttps://events.icecube.wisc.edu/event/141/co ntributions/7975/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7975/ END:VEVENT BEGIN:VEVENT SUMMARY:Deep learning in astroparticle physics DTSTART:20220201T141500Z DTEND:20220201T150000Z DTSTAMP:20260722T082500Z UID:indico-contribution-8056@events.icecube.wisc.edu DESCRIPTION:Speakers: Jonas Glombitza (RWTH AACHEN UNIVERSITY)\n\nIn the p ast few years\, deep-learning-based algorithms have been extraordinarily s uccessful across many domains\, including computer vision\, machine transl ation\, engineering\, and science. Also\, in physics\, applications are ac cumulating due to the need for fast and precise algorithms that are able t o exploit huge amounts of data. So\, could it even become a new paradigm f or data-driven knowledge discovery?\n\nIn this contribution\, we introduce the fundamental concepts of deep learning\, review the potential of this emerging technology\, and illustrate the wide variety of possible applicat ions in the context of particle and astroparticle physics.\nFinally\, we p resent novel approaches in the field and discuss future applications.\n\nh ttps://events.icecube.wisc.edu/event/141/contributions/8056/ LOCATION:Embassy Suites by Hilton Newark Wilmington South URL:https://events.icecube.wisc.edu/event/141/contributions/8056/ END:VEVENT BEGIN:VEVENT SUMMARY:Workshop on Machine learning for Cosmic-Ray Air Showers - Summary & Outlook DTSTART:20220203T215500Z DTEND:20220203T222000Z DTSTAMP:20260722T082500Z UID:indico-contribution-8058@events.icecube.wisc.edu DESCRIPTION:Speakers: Matthias Plum (Marquette University)\n\nSummary of t he 3-day workshop and outlook into the future of cosmic rays analysis usin g state of the art machine learning techniques.\n\nhttps://events.icecube. wisc.edu/event/141/contributions/8058/ LOCATION:Embassy Suites by Hilton Newark Wilmington South URL:https://events.icecube.wisc.edu/event/141/contributions/8058/ END:VEVENT BEGIN:VEVENT SUMMARY:Machine learning in Baikal-GVD DTSTART:20220203T140000Z DTEND:20220203T143500Z DTSTAMP:20260722T082500Z UID:indico-contribution-7990@events.icecube.wisc.edu DESCRIPTION:Speakers: Ivan Kharuk (Institute for Nuclear Research RAS)\, f or Baikal-GVD collaboration\, Albert Matseyko (Institute for Nuclear Resea rch RAS / Moscow Institute for Physics and Technology)\n\nBaikal-GVD is a large-scale underwater neutrino telescope currently under construction in Lake Baikal. Its principal component is a three-dimensional array of optic al modules (OMs) registering Cherenkov light associated with the neutrino- induced particles. The OMs are organized in clusters\, each containing 8 v ertical strings with 36 OMs per string.\n\nLocated in a natural water rese rvoir\, the OMs are exposed to the luminescence of the Baikal water. This necessitates the search for highly effective algorithms for noise rejectio n as the first step of data analysis. We developed a convolutional neural network reaching ~97% signal purity (precision) and ~99% survival efficien cy (recall) for the signal hits on Monte-Carlo data. The architecture of t he neural network exploits the causal connection between individual hits\, rather than their spatial location.\n \nThe other problem we are solving with the help of neural networks is a reliable identification of neutrino events. The underlying issue is that muons flux due to cosmic rays is many orders of magnitude higher than that of neutrinos. Hence the discriminati ng algorithm must have extremly small error rate. We discuss how this can be achieved by adjusting event weights and choosing a proper loss function for the neural network.\n\nhttps://events.icecube.wisc.edu/event/141/cont ributions/7990/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7990/ END:VEVENT BEGIN:VEVENT SUMMARY:Machine Learning and Artificial Intelligence in Physics: Overview and Applications DTSTART:20220202T210000Z DTEND:20220202T223000Z DTSTAMP:20260722T082500Z UID:indico-contribution-8057@events.icecube.wisc.edu DESCRIPTION:Speakers: Gregory Dobler (University of Delaware)\n\nThe use o f computational algorithms\, implemented on a computer\, to extract inform ation from data has a history that dates back to at least the middle of th e 20th century. However\, the confluence of three recent developments has led to rapid advancements in this methodology over the past 15-20 years: the advent of the era of large datasets in which massive of amounts of dat a can be collected\, stored\, and accessed efficiently\; the development o f computational algorithms that can perform classification and prediction to high degrees of accuracy across a variety of applied situations\; and b road access to the computational power of modern computing systems that al low for the building of complex models of phenomenology in diverse domains . In this talk I will describe the basic fundamentals of Machine Learning (ML)\, how ML is used to extract information from data\, the potential pi tfalls to avoid when using ML in a variety of applications\, the relations hip between ML and what is currently commonly referred to as Artificial In telligence (AI)\, and the transferability of ML from physics-based to non physics-based problems. The second half of this presentation will consist of a live demo applying ML to a physics application in the Python coding l anguage using publicly available tools.\n\nhttps://events.icecube.wisc.edu /event/141/contributions/8057/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/8057/ END:VEVENT BEGIN:VEVENT SUMMARY:Training Neural Networks to Classify and Denoise Cosmic-Ray Radio Signals Using Background Measured at the South Pole DTSTART:20220203T201500Z DTEND:20220203T203000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7981@events.icecube.wisc.edu DESCRIPTION:Speakers: Alan Coleman (University of Delaware)\, Dana Kullgre n (University of Delaware)\, Frank Schroeder (University of Delaware / Kar lsruhe Institute of Technology)\, IceCube Collaboration\, Abdul Rehman (Un iversity of Delaware)\n\nCosmic-ray air showers produce radio signals whic h can be detected from Earth’s surface. However\, the radio background t hat is detected along with these signals can make it difficult to identify an air shower signal from the local background. To solve this problem\, t his project aims to train two convolutional neural networks (CNNs): a “c lassifier” and a “denoiser”. The classifier distinguishes a trace co ntaining an air shower signal from a trace containing only background. The denoiser takes a noisy signal and removes the noise (background) from it. The dataset used to train these networks includes simulated air shower si gnals produced in CoREAS as well as background traces recorded with a prot otype station at the IceCube Neutrino Observatory at the geographic South Pole. The training and analysis is performed using the frequency band from 100 to 350 MHz. The goal of these CNNs is to improve the detection thresh old of radio experiments to detect signals with lower energies and to impr ove the removal of background noise from air shower radio signals. I will show how the CNNs perform in identifying cosmic ray signals and in extract ing air shower pulses from the noisy waveforms.\n\nhttps://events.icecube. wisc.edu/event/141/contributions/7981/ LOCATION:Embassy Suites by Hilton Newark Wilmington South URL:https://events.icecube.wisc.edu/event/141/contributions/7981/ END:VEVENT BEGIN:VEVENT SUMMARY:Measurement of the high-energy muon multiplicity in cosmic-ray air showers with IceTop and IceCube using neural networks DTSTART:20220201T190000Z DTEND:20220201T193000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7996@events.icecube.wisc.edu DESCRIPTION:Speakers: Stef Verpoest (University of Gent)\, The IceCube Col laboration\n\nThe IceTop and IceCube detectors at the South Pole provide t he opportunity to simultaneously measure the electromagnetic and low-energ y muonic component of a cosmic-ray air shower at the surface\, and the pen etrating muons in the deep ice. Various properties of the bundle of muons above several 100 GeV measured in IceCube are sensitive to the mass of the primary cosmic ray and contain information about the hadronic physics of the first interactions in the atmosphere. By combining a maximum-likelihoo d reconstruction of the energy loss of the muon bundle with a simple Convo lutional or Recurrent Neural Network\, the multiplicity of muons above a c ertain energy threshold in the shower can be estimated with reasonable acc uracy. Along with information on the electromagnetic shower component as m easured by IceTop\, this opens the possibility for a measurement of the ev olution of the average high-energy muon content of air showers with primar y energies from PeV to EeV.\n\nhttps://events.icecube.wisc.edu/event/141/c ontributions/7996/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/7996/ END:VEVENT BEGIN:VEVENT SUMMARY:Composition of 100 TeV - 100 PeV Cosmic Rays with IceCube and IceT op using Boosted Decision Trees DTSTART:20220201T210000Z DTEND:20220201T213000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7978@events.icecube.wisc.edu DESCRIPTION:Speakers: Julian Saffer (Karlsruhe Institute of Technology)\, IceCube Collaboration\n\nIceTop is the surface component of the IceCube So uth Pole Neutrino Observatory and dedicated to the indirect detection of c osmic rays (CRs). The recent implementation of a new trigger that only req uires 2 of IceTop's 6 central infill stations hit by a CR-induced air show er allowed to reduce the primary energy threshold for the detection of low -energy CRs from 1.6 PeV to 250 TeV. This lead to a narrowing of the gap b etween direct and indirect CR measurements and coverage of the entire knee region of the spectrum.\n\nApart from the reconstruction of primary energ y\, shower core position and zenith angle\, this work aims to create a sup ervised machine-learning model that is capable of correctly predicting the mass composition of CR primaries. This requires the combination of signal s from the surface and the corresponding tracks of high-energetic muons wi thin the deep in-ice detector below. For this purpose\, tree-based methods \, namely random forests and boosted decision trees\, have been trained fo r regression and classification tasks on Monte Carlo shower data of four p rimary types. Additionally\, plans for a potential implementation with neu ral networks are presented.\n\nhttps://events.icecube.wisc.edu/event/141/c ontributions/7978/ LOCATION:Embassy Suites by Hilton Newark Wilmington South URL:https://events.icecube.wisc.edu/event/141/contributions/7978/ END:VEVENT BEGIN:VEVENT SUMMARY:Extraction of the Muon Signals Recorded with the Surface Detector of the Pierre Auger Observatory Using Recurrent Neural Networks DTSTART:20220202T163000Z DTEND:20220202T170000Z DTSTAMP:20260722T082500Z UID:indico-contribution-8059@events.icecube.wisc.edu DESCRIPTION:Speakers: Juan Miguel Carceller (University College London)\n\ nWe present a method based on the use of Recurrent Neural Networks to extr act the muon component from the time traces registered with water-Cherenko v detector (WCD) stations of the Surface Detector of the Pierre Auger Obse rvatory. With the current design of the WCDs it is not straightforward to separate the contribution of muons to the time traces from those of photon s\, electrons and positrons in cosmic ray showers dominated by electromagn etic particles. Separating the muon and electromagnetic components is cruc ial for determining the nature of the primary cosmic ray and properties of hadronic interactions at ultra-high energies. We trained the neural netwo rk to extract the muon and the electromagnetic components from the WCD tra ces using a large set of simulated air showers\, with energies between $10 ^{18.5}$ eV and $10^{20}$ eV and zenith angles below 60 degrees. The perfo rmance of this method is studied on experimental data of the Pierre Auger Observatory. It is shown that the predicted muon lateral distributions agr ee with the parameterizations obtained by the AGASA collaboration.\n\nhttp s://events.icecube.wisc.edu/event/141/contributions/8059/ LOCATION:Online URL:https://events.icecube.wisc.edu/event/141/contributions/8059/ END:VEVENT BEGIN:VEVENT SUMMARY:Neural Network Approaches for Event Classification Onboard EUSO-SP B2 DTSTART:20220202T170000Z DTEND:20220202T173000Z DTSTAMP:20260722T082500Z UID:indico-contribution-7979@events.icecube.wisc.edu DESCRIPTION:Speakers: George Filippatos (Colorado School of Mines)\n\nThe Extreme Universe Space Observatory Super Pressure Balloon 2 (EUSO-SPB2) is under development\, and will prototype instrumentation for future satelli te-based missions\, including the Probe of Extreme Multi-Messenger Astroph ysics (POEMMA). EUSO-SPB2 will consist of two telescopes. The first is a C herenkov telescope (CT) being developed to identify and estimate the backg round sources for future below-the-limb very high energy (E>10 PeV) astrop hysical neutrino observations. The second is a fluorescence telescope (FT) being developed for detection of Ultra High Energy Cosmic Rays (UHECRs). \n\nSuper pressure balloons (SPB) are inherently risky due to the lack of flight controls compared to other orbital and suborbital crafts. The recov ery of data from the instrument is only possible if the mission is termina ted over land\, therefore the only guaranteed data is what can be download ed during the flight. Limited satellite based telemetry being shared betwe en the two telescopes and housekeeping data results in roughly 1% of event s recorded with the FT being downloaded during the flight. This necessitat es onboard classification schemes to assign priority to data to be downloa ded\, which can be run using the limited computational resources of the SP B. We implement several architectures to achieve classification including convolutional\, recurrent and Long Short Term Memory (LSTM) neural network s. These networks were trained using a large library of simulated EAS sign als and both simulated noise and data taken from previous EUSO experiments . Ultimately\, the neural network approach shows great promise but will re quire additional pre-flight testing in order to be fully validated.\n\nhtt ps://events.icecube.wisc.edu/event/141/contributions/7979/ LOCATION:Embassy Suites by Hilton Newark Wilmington South URL:https://events.icecube.wisc.edu/event/141/contributions/7979/ END:VEVENT END:VCALENDAR